RAM test

[zacl87]

Hey so the over the last week or so I've slowly figured out that the constant kernel panics that have prevented me to sometimes even start up have been from a malfunctioning ram card that was obviously dragging everything else with it when it failed.

I finally figured it out by doing the hardware test a few times (first couple of times it didn't show up - ???) and it showed it was a memory failure. I reseated the chips and no dice so I took them in/out and swapped them etc until I found out which one was causing trouble.

I'm now running on just 1 card (1gb) awaiting the new one I ordered. The question I guess I want to know is are there any RAM tests you can do periodically that can show a card that's failing?

Looking back I now realise that there were probably more beachballs/unexpected quitting than usual before it died. Is there any way to prevent it crashing out of the blue by realising it's slowly dying?

It seems like something that should show a warning or something before it completely fails?

 

[BlueRevolution]

Why? This isn't a hard drive. It's not like you lose data or anything when it fails. That's why there's no SMART analogue. There's no preventive maintenance you can perform to make it last longer. Memory is just one component that you look at when a problem does occur, as it did in your case.

 

[zacl87]

Well it's hard to tell if something's failing. More beachballs than usual doesn't necessarily mean your ram is dying. I'd much prefer to know it's running at a lower capacity than it should so I could replace it.

What's the point of having something run badly until it dies? It's like having a screen with a quarter of dead pixles and just leaving it until the entire screen dies.

I don't know maybe i'm just picky. It seem annoying to me though.

 

[dolphin842]

Short of running a thorough memory test from time to time, as BlueRevolution says, I'm not aware of any way to actively 'monitor' the health of RAM.

From what I understand, the process of RAM testing involves continuous writing/reading of data to/from the RAM modules. In turn, it would be difficult to make a program that did this 'in the background,' as it would exact a significant performance penalty (and because the computer would be in use, the test wouldn't be able to reach all of the RAM anyway).

 

[BlueRevolution]

Well it's hard to tell if something's failing. More beachballs than usual doesn't necessarily mean your ram is dying. I'd much prefer to know it's running at a lower capacity than it should so I could replace it.

What's the point of having something run badly until it dies? It's like having a screen with a quarter of dead pixles and just leaving it until the entire screen dies.

I don't know maybe i'm just picky. It seem annoying to me though.

Actually, it's like having a screen in perfect condition, and running periodic tests so you can find out if it's about to get a bad pixel. There's no point.

 

[zacl87]

Actually, it's like having a screen in perfect condition, and running periodic tests so you can find out if it's about to get a bad pixel. There's no point.

Well, I'm obviously not referring to ram that's in perfect condition.. A simple no will suffice next time.